[sdiy] Digital Noise, FYI

[Noah Vawter]

On Feb 17, 2011, at 3:02 PM, Scott Gravenhorst wrote:

> Matthew Smith <matt at smiffytech.com> wrote:
>> Quoth Scott Gravenhorst at 16/02/11 23:16...
>> ...
>>> The list includes LFSR lengths from 3 bits to 168 bits.  The more  
>>> taps, the better.  I
>>> typically use 64.
>
> OOPS - I meant - "the more bits the better" not "the more taps the  
> better" ...

Yes, interestingly, I've seen a chart of the necessary taps up to 144  
bits elsewhere, and it's amazing how the most taps it ever requires  
(for a maximal-length) is 4, and many of them only require 2.  That's  
just sort of meta-physically amazing.



>> The CPLD I'll be using won't be performing any other functions - I  
>> think
>> I'll just keep adding stages until the ISE software tells me I've run
>> out of resources!
>>
>> I'm guessing that I'd feed my 12-bit DAC from the lower 12 stages -  
>> is
>> there a better point or is this OK?
>>
>> Will probably have my clock range from about 1kHz to 1MHz - upper end
>> for noises, lower end for Weird Noise :-)
>
> It makes no real difference what 12 bits you pick.  I use the ones  
> on the low end, but the high
> end, in the middle - won't change what it sounds like.

I've started a project to characterize the various LFSR patterns  
sonically...     but I have this pesky thing called a dissertation  
that is screwing my life atm so I can't finish it yet.  Look for the  
LFSR stuff after June :)